FDA's AI Integration Faces Leadership Uncertainty After High-Level Departures
核心洞察
The FDA's AI use-cases surged 148% between 2024 and 2025, driven by former commissioner Marty Makary's centralized approach and the rollout of the agency-wide Elsa (搜索) large language model tool.
Recent departures of Makary, chief AI officer Jeremy Walsh, and acting CIO Sridhar Mantha have created uncertainty around leadership and governance of FDA-wide AI efforts.
Former FDA AI policy official Tala Fakhouri warns the agency may revert to a fragmented, department-specific AI approach, though sponsor-facing AI policies remain unchanged.
The FDA's ambitious push to integrate artificial intelligence across its operations faces an uncertain trajectory following the departure of several key leaders, including former commissioner Marty Makary, who championed a centralized AI strategy for the agency.
Between 2024 and 2025, the number of AI use-cases reported by the FDA skyrocketed by 148%, according to a Bipartisan Policy Center study, reflecting a broader government-wide push to leverage the technology. Under Makary's leadership, the agency rolled out Elsa (搜索), an agency-wide large language model designed to accelerate reviews and evaluations by handling support tasks such as writing and summarizing reports. In December, the FDA also announced plans to expand the use of agentic AI to support premarket reviews, inspections, and administrative tasks.
Acting commissioner Kyle Diamantas has stated that AI remains a top priority for the FDA. However, the recent high-level departures — including Makary, Jeremy Walsh, the chief artificial intelligence officer, and Sridhar Mantha, the acting chief information officer — have raised questions about the future of AI implementation at the agency.
"It's unclear now what the leadership and governance structure are around FDA-wide efforts," said Tala Fakhouri, chief artificial intelligence and regulatory strategy officer at Parexel and a former FDA AI policy official.
Signs of Fragmentation
Fakhouri noted that while FDA officials have spoken about AI at recent conferences, they were all representing individual divisions rather than presenting a unified agency-wide vision. This, she suggested, could portend a return to the more fragmented, department-specific AI approach that existed prior to Makary's centralized push. Efforts to increase transparency around AI use and enact AI-related policy-making could also slow.
These potential setbacks primarily concern the agency's internal use of AI, not the policies governing how drug companies apply the technology in their own work. So far, there has been no change in the FDA's policies as they relate to sponsor use of AI. "And I don't expect to see a change there, which is good," Fakhouri said.
The Evolution of Elsa (搜索)
Elsa (搜索) had its origins in a CDER-developed program called CDER GPT, according to Fakhouri. The agency expanded on that original program by incorporating a retrieval-augmented generation system designed to reduce AI hallucinations. The system confines the large language model to a well-defined database of trusted information, tailored for the individual centers within the agency.
This architecture allows staff to use the tool for role-specific tasks. For instance, staff members can quickly generate a summary of industry comments on a particular proposal, or the Office of New Drugs could use Elsa (搜索) to rapidly compile a history of regulatory submissions spanning years. While Elsa has been leveraged for these types of burdensome tasks, Fakhouri does not believe the technology is used for final decision-making. Still, "staff can use the tools to augment the work that they're doing. We should all be happy about that," she said.
Calls for Transparency and Streamlined Rulemaking
Fakhouri emphasized that transparency about how the FDA is using AI in its processes remains lacking. "If the regulators are using AI in certain ways to augment reviewer work or to become an assistant to a reviewer, I think it's good practice for industry to know what these uses look like," she said.
Greater transparency could also foster collaboration with industry. "If I am a sponsor or a CRO preparing a submission on behalf of a sponsor, I can prepare my submission with all the data labels and information that might be helpful for the reviewer and the [AI assistant] to be able to review my package," Fakhouri explained. She expects the agency to move toward greater transparency over time, but added, "it would require someone in a leadership position at the agency to become aware of that and want to actually make it happen."
Fakhouri also called for the agency to streamline AI-related rulemaking that impacts the industry, such as how AI tools are validated for clinical trials. Under Makary, policy changes were sometimes announced outside the traditional FDA guidance process — through journal articles or press conferences instead. The agency now appears to be returning to its former norms under Diamantas, who recently confirmed that informal statements made by the former commissioner do not represent official policy.
"They will go through the regular guidance and policy development processes," Fakhouri said. However, she acknowledged the tension between structure and speed: "The fastest guidance that you could put out would still probably take a year. A year in the age of AI is very slow."
Finding a balance between structure and flexibility, Fakhouri concluded, will be one of the FDA's key policy challenges going forward.